Coupled Cluster methods such as CCSD and CCSD(T) accurately predict molecular geometries and properties at equilibrium and therefore are considered as ‘gold-standard’ for all ab-initio and DFT based methods. But there are situations in which these methods fail, especially when the single reference approximation is no longer valid, when there are more configurations that need to be included at the same scale as the starting reference. A CC or CI based multi-reference approach is generally advocated and has been shown to yield precise results in such situations. But such calculations are computationally expensive and hardly even practical for systems of modest size. A current intense topic of research is how, if at all, we can describe these ‘multi-reference’ situations through a single reference formulation where an approximate effect of higher excitations (quadruples, pentuples and so on) is included by deleting some terms from the general CCSD equations. This is based on the hypothesis that some terms present at lower level CC methods such as CCD, CCSD will eventually cancel out with other terms present in methods as CCSDTQ, CCSDTQP and so on. The idea here is then to select these terms and remove them even for lower level CC calculations. A new approach named Distinguishable Cluster approach, an approximation of the general CC method, performs very well in these troublesome quasi-degenerate situations. We try to rationalize why DCC and other methods, generally categorized as approximate Coupled-Cluster methods, seem to give improved results even though they are obtained by deleting some terms from the exact formulation of CC methods.

Approximate coupled-cluster methods: Addition by subtraction?

251st National Meeting (2016)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.